November 2015 was warmest November on record | Hacker News Reader

November 2015 was warmest November on record

ncdc.noaa.gov

50 pointscryptoz27 comments

So I'll take one of the statements to ask a question.

> This was the highest for November in the 1880–2015 record, surpassing the previous record set in 2013 by 0.27°F (0.15°C)

So the NOAA claims that this November was 0.15C higher than the previous record high.

What does this mean in a larger climate context?

Probability theory tells me that a single number like this is meaningless, but other than that I don't know how to interpret this except that it is information that was published with no specific agenda other than reporting it.

It seems the number of records set per year in either direction is a useful signal to climatologists. More records than prior years indicates a trend.

> Probability theory tells me that a single number like this is meaningless

That is true. But when considering that the ten warmest years since 1880 all have occurred since 1998[1] it is no longer a single number.

I'm guessing after 2015 that the eleven warmest years will have occurred after 1998.

[1] http://climate.nasa.gov/blog/2224

I'm not sure that the single number is entirely meaningless, because you would expect reversion to the mean in the absence of a trend, so if you computed the likelihood function then I guess it would be weighted toward there being a positive trend, if only slightly. I think :-/.

The null hypothesis (no warming) would posit independent events with equal probability (the chance of an above-average and below-average year are equivalent).

So in that case you're basically asking the question "given that I've flipped a fair coin and gotten heads 9 times in a row, what is the probability of heads on the 10th flip?" It's deceptive because you want to say 0.5^10, but the answer is actually 50% (because it's one throw of a fair coin). The 0.5^10 is the chance of 10 heads in a row occurring, but you can't disregard the prior (you have already thrown 9 heads).

Reversion to the mean is a trend, it says nothing about any individual event in a series. In the event that it's not warming, the chance of a warm year is just as likely as a cool year.

Now, as a trend, the evidence is highly suggestive of global warming being the case. When you have a sufficient sample of throws with a huge number of heads, the chance of the coin being biased* increases.

* Note: you cannot actually bias a flipped coin. But you can bias a die, or a spun coin: https://www.stat.berkeley.edu/~nolan/Papers/dice.pdf

>"The null hypothesis would posit independent events with equal probability (the chance of an above-average and below-average year are equivalent)."

Why should that be the null hypothesis? Is that a theory someone subscribes to? That would be a theory of no trends and no auto-correlation in temperature. I doubt you will find someone espousing such a theory, so why should we make something no one believes a null hypothesis? If you can find someone claiming that, I am interested in the reasoning.

I think the difference in this case is that we are looking at a point that exceeds the current series maximum. If we had a single value from a series without a trend, we would expect it to fall below the maximum value observed so far (perhaps after sampling at least 2 points - I will think about that later), and so I reckon that the likelihood of a positive trend will be higher than for a zero or negative trend because of that fact.

I may try and actually work it out to see if what I'm saying is correct, but one thing I can tell you that likelihood of me not doing so is greater ;-).

Yes, looking at the sample minimum/maximum does change the probabilities. You're looking for Extreme Value Theory.

https://en.wikipedia.org/wiki/Extreme_value_theory

I kept it simple (above or below average should be equal probability if your average is really average) to try and avoid that. I have no idea what kind of distributions weather follows, and as the other guy noted weather is not fully independent either. But it works for a simple explanation. :)

It's been a long time since my stat courses and I don't remember this stuff all that well. Maybe there is something that expresses it better. Either way the fact that we're suddenly having a big run of Nth Warmest Year events is pretty suggestive that something is going on. But you do have to look at it as a trend, particular years aren't all that meaningful.

Two things strike me about that quote.

First, the range starts in 1880, which is just about the end of the Little Ice Age.[1] So yeah, pretty much every year since then is likely to be warmer than the last one. For a fairer comparison, you'd have to go back to 1100 or so to the Medieval Warm Period. We're still warmer today than back then, but not by as much.

The second thing that struck me was the same thing that struck you: a single number is meaningless. If the fluctuations were completely random, you'd still always have a 'highest' one, but it wouldn't mean anything. And since we're trending upwards anyway (still coming out of that cool period, which only ended 100 or so years ago) you'd expect new highs to occur fairly often and fairly recently.

I don't think this particular record says much; at most it fits with the fairly obvious warming trend we're in. If it was the coldest November since 1880, that'd be news because it would be a really odd occurrence.

[1] https://en.wikipedia.org/wiki/Little_Ice_Age

Here is also an article by the Capital Weather Gang: https://www.washingtonpost.com/news/capital-weather-gang/wp/...

There is a lot of agreement, even among the anti-CAGW people, that we have had a gradual trend towards higher global temperatures for many decades, and longer if you are OK not using the satellite record. Looks like that trend may be continuing. It's definitely at least a warm end of the year for the eastern US.

More anecdotal than anything, but this is the warmest December in Boston that I've ever seen. It's 50 outside right now.

I've not been in Houston long (6 years) but this December also has been much warmer than I've experienced.

Anecdata ftw!

We've been in the 60's and 70's for all of December so far here in OKC. I've lived here my whole life and it usually doesn't get above 40-50 this time of year.

We've hit 70's in January often enough over the years.

Yup, NYC here, been here 20+ years, warmest December I can remember. It was 60 degrees last weekend!

Anecdotally, where I live, this December is cooler than last year's. But I've only lived here 1.5 years.

This is more likely El Nino-related (the extremely strong El Nino currently in process is likely to make the East & South much warmer than usual), though of course worldwide temperature increases will compound to make the temperature higher than it might have been without climate change. My understanding is that it's still fairly unclear how climate change might be affecting equatorial oscillations like El Nino (as opposed to the Arctic/North Atlantic oscillations where there are some pretty specific mechanisms of action hypothesized).

It definitely is… but what's the extremely strong El Niño related to?

Isn't it glorious?!!?? But don't get fooled- last year was quite warm also, until....

BTW, "swampthinker" makes me think you live here on the North Shore.

Los Angeles has been freezing... at 60 or less, haha.

Not according to the RSS and UAH satellite tropospheric data sets.

The MSU and AMSU sensors have issues, but still have wider coverage and superior accuracy to the ground station data the NOAA is using. Additionally, RSS and UAH are two different teams, with different methodologies, and using different sensors, with corroborating data.

http://images.remss.com/msu/msu_time_series.html

http://vortex.nsstc.uah.edu/data/msu/v6.0beta/tlt/tltglhmam_...

[EDIT] I'm absolutely wrong. Both the RSS and UAH data confirm that November is the hottest November in the sattelite record also.

Thanks jacobolus for pointing this out (and making sure I stop assuming the NOAA is exaggerating on its hottest month claims.)

Here are some of the relevant rows from your second link (hottest values in one of the “GLOBAL” columns). I’m not sure I completely understand what the columns all mean though. Can you elaborate a bit?

   YEAR  MON  GLOBAL     NH      SH    TRPC  NO.DAYS   GLOBAL     NH      SH    TRPC  DAYS
   1990   11   0.255   0.268   0.242   0.121    0      -0.005  -0.037   0.026  -0.092    0
   1998   11   0.121   0.126   0.116   0.111    0       0.484   0.502   0.465   0.733    0
   2001   11   0.168   0.203   0.133   0.153    0       0.101   0.108   0.095   0.013    0
   2002   11   0.196   0.164   0.227   0.242    0       0.215   0.172   0.257   0.192    0
   2005   11   0.198   0.234   0.161   0.209    0       0.197   0.240   0.154   0.293    0
   2009   11   0.277   0.203   0.350   0.428    0       0.093   0.091   0.094   0.122    0
   2010   11   0.170   0.260   0.080  -0.202    0       0.343   0.405   0.281   0.430    0
   2014   11   0.229   0.251   0.207   0.239    0       0.178   0.203   0.153   0.127    0
   2015   11   0.330   0.433   0.228   0.525    0       0.248   0.343   0.153   0.314    0

Maybe you can also extract and quote the relevant data from your first link? I’m not sure what I should be looking at.

Wow, I totally missed that. November was incredibly warm. This will bring up the long term trends significantly.

Looks like you excerpted the UAH data which is year, month, global average, Northern Hemisphere, Southern Hemisphere, Tropics, not sure what No.Days is for, and then the 12 month running mean for each of the previous.

I'm just so used to the NOAA saying "Warmest month ever" and it being contradicted by the MSU/AMSU data, it didn't even occur to me they were right.

Thank you so much for double checking me.

According to that then it looks like the 12 month running average is significantly lower, is that to be expected/in line with predictions/at all statistically relevant? I'm still not sure I actually understand what those numbers represent.

EDIT: After jacobolus updated with more columns and checking some other years from the link provided, that does not actually appear to be a trend. My apologies.

Here are the 27 hottest months on record, if anyone was curious about comparing to e.g. the hottest october or december:

   YEAR  MON  GLOBAL     NH      SH    TRPC 
   2010    4   0.330   0.449   0.210   0.687
   2015    6   0.330   0.410   0.251   0.456
   2015   11   0.330   0.433   0.228   0.525
   2004    3   0.334   0.490   0.179   0.271
   2010    6   0.336   0.432   0.240   0.449
   2010    7   0.354   0.520   0.188   0.315
   2003   12   0.357   0.462   0.252   0.272
   2010    8   0.370   0.518   0.222   0.272
   1987   12   0.382   0.573   0.191   0.656
   2010    9   0.393   0.356   0.429   0.144
   1998   10   0.400   0.484   0.315   0.368
   2007    1   0.406   0.484   0.327   0.550
   2010    5   0.426   0.611   0.241   0.804
   2015   10   0.426   0.638   0.213   0.530
   2013    1   0.429   0.345   0.512   0.420
   1998    9   0.437   0.506   0.367   0.356
   2010    2   0.441   0.392   0.491   0.856
   1998    3   0.482   0.509   0.455   1.128
   2010    1   0.483   0.553   0.412   0.599
   1998    1   0.491   0.402   0.580   1.132
   1998    7   0.496   0.617   0.375   0.467
   2010    3   0.503   0.560   0.446   0.776
   1998    8   0.513   0.482   0.544   0.459
   1998    6   0.561   0.572   0.551   0.629
   1998    5   0.636   0.566   0.706   0.975
   1998    2   0.665   0.688   0.642   1.275
   1998    4   0.742   0.852   0.633   1.177

1998 was just incredibly hot throughout (but not in November).

http://www.npr.org/2014/01/07/260408972/polar-vortex-blames-...

http://www.npr.org/2015/12/15/459871063/climate-change-plays...

two NPR links here, the 2014 says how the cold in winter 2014 is the polar vortex, and the warm in 2015 is a combination of tight polar vortex and el nino, he* resisted saying it was related to climate change.

he is Mike Halpert, NOAA's deputy director of the Climate Prediction Center

Reply on news.ycombinator.com